From time to time, I’ll get a scanned pdf/image of a document to translate, in order to translate quickly, I will often try to convert all the text contents to editable text or .doc, .docx formats. This helps a lot in researching the term, especially for Chinese to English, as in the Chinese language, there are no spaces to separate the characters to clearly indicate which one forms a noun, which is a verb and when I come across a string which I need to ascertain the noun, I need to spend time researching on the Internet.

As such, I have tried several OCR (Optical Character Recognition) Software. The evaluation below is based on a few scanned .pdf files of articles (with both English words and Chinese characters) that I have tried to convert to text/.doc files.

1. Adobe Acrobat Pro

While this does convert all the .pdf to .doc files, actually what I got was several small images in each .doc file, and not editable text.

2. Free Online OCR (http://www.newocr.com/)

This actually extracts the text and has very good accuracy compared to the others I have tried. But the accuracy reduces the larger the area or the more contents you want to convert. Accuracy is quite poor for a full page of Chinese characters, but for a single paragraph of about 200 chars, it works pretty well.

It support multiple languages, which is another plus point, and no registration required if you just want to download the text (after conversion)

3. docs.zone (https://docs.zone/ocr)

I did not really try this as it only supported certain European languages (English, French, German, Italian, Portuguese, Spanish, Russian). But it claims that the OCR conversion speed is approximately one page per 20-30 seconds. You need to register if you want the whole file.

4. Free Online OCR Services (https://www.onlineocr.net/)

Another one with pretty good accuracy but you need to sign up to access the full range of services. I have only tried the guest mode which will only convert one page or less. Problem with this is that it does not work with all the .pdf files I have. For two files, I got an invalid content error. For the files with no error, the conversion is quite accurate for both English words and Chinese characters, I only have to edit about 25% of the text.

5. Another free online OCR (http://www.free-online-ocr.com/)

I only tried one .pdf document (4 pages, dual columns, mixed of Chinese and English) and I got the same result for Adobe Acrobat 10, except that I got even smaller images, which I then extracted and used the online free OCR (one image at a time) to get the text extracted. Maybe I need to explore more.

Conclusion

My first choice will be 4. and then 2. However, I am still looking for a more efficient way to convert as these still require a lot of time, and I need to break down the file into many pieces to get a reasonable accuracy.